5b09374c77
getting training and testing loop working, adding tqdm to simulator, and reduced number of frames, updating readme
Arec
2021-07-23 04:44:50 -07:00
8a4094037a
update forward pass test
Johannes Fischer
2021-07-23 11:16:08 +02:00
acd2b730a6
adding transforms to test case
Arec
2021-07-23 01:38:20 -07:00
9c7fc47abe
updating test batch
Arec
2021-07-23 01:27:57 -07:00
5d3b3fa1bc
Merge branch 'main' of github.com:sisl/InteractionImitation
Johannes Fischer
2021-07-23 09:30:26 +02:00
9b0aeb182f
Add .vscode to gitignore
Johannes Fischer
2021-07-23 09:30:23 +02:00
563a2cfbd4
making a differentiable transform for use for pytorch, making sure the fitting function treats nans properly while fitting. next issue: forward pass is returning nans
Arec
2021-07-22 12:37:36 -07:00
1c22bd6111
bugfix in deepsets
Johannes Fischer
2021-07-22 18:44:12 +02:00
91d052445e
making test case for typing bug and fixing some small typing errors in bc
Arec
2021-07-22 08:23:46 -07:00
827a8e7172
bugfix in Phi module
Johannes Fischer
2021-07-22 14:33:05 +02:00
1ca9914bf9
fixing bugs in transform, expert demo processing, main train function, and behavior cloning class. need to get bc class parameters to return nonempty list
Arec
2021-07-21 09:44:20 -07:00
18b8e0c58f
updating main testing function to use configs and seeds, finishing first pass at behavior cloning policy and training loop. not yet tested
Arec
2021-07-21 08:22:08 -07:00
4350cf8cd5
defining transform class to do tensor size manipulation before and after transform
Arec
2021-07-21 08:20:55 -07:00
3422e9c9ef
exporting policy class and letting default final activation do nothing
Arec
2021-07-21 08:20:15 -07:00
c939a8e131
Remove todo
Johannes Fischer
2021-07-20 18:45:49 +02:00
ca0f520c89
Merge branch 'main' of github.com:sisl/InteractionImitation
Johannes Fischer
2021-07-20 18:44:11 +02:00
61f06c95c2
Improve deepsets module
Johannes Fischer
2021-07-20 18:44:07 +02:00
ffdff12ccb
Update deepsets to deal with nans (first version)
Johannes Fischer
2021-07-20 18:23:51 +02:00
6794b4cad8
making saving and loading functions class requirements, working on behavior cloning policy class and training function
Arec
2021-07-20 08:29:00 -07:00
1a74fa5237
making general-purpose metric function
Arec
2021-07-20 07:05:09 -07:00
226a427436
Adapt policy input format
Johannes Fischer
2021-07-20 15:45:15 +02:00
7109afb21f
Merge branch 'main' of github.com:sisl/InteractionImitation
Johannes Fischer
2021-07-20 15:25:10 +02:00
8534e8a437
Update scratch space
Johannes Fischer
2021-07-20 15:25:07 +02:00
ba10a1f53b
Implement transforms for normalization
Johannes Fischer
2021-07-20 15:24:34 +02:00
a15e8c29ff
moving transforms out of dataset class, will be exclusively in policy classes
Arec
2021-07-20 06:16:54 -07:00
2fb5d5e5b1
developing main experiment loop, functions required to implement in bc and other imitation methods
Arec
2021-07-20 05:58:24 -07:00
8acfbc7993
updating expert data to keep track of deviations and check for collisions
Arec
2021-07-19 07:10:28 -07:00
8103a98377
Add .project to gitignore
Johannes Fischer
2021-07-19 12:36:01 +02:00
e2aac7424f
Add requirement
Johannes Fischer
2021-07-19 12:29:20 +02:00
2373d924b7
adding expert data loader and saver. same issue as previous commit, unable to get through full simulation with backtracked accelerations
Arec
2021-07-16 08:33:40 -07:00
a6e691bd1a
working on expert data saver and loader. mostly working, but issue getting all the way through simulation with backtracked accelerations
Arec
2021-07-16 08:32:48 -07:00
b684a5db6d
making initial file shell
Arec
2021-07-16 07:15:33 -07:00